Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Table top computer

Author: Will Singleton

Date: 13:09:56 07/08/99

Go up one level in this thread



On July 08, 1999 at 13:51:36, Christophe Theron wrote:

>On July 07, 1999 at 14:10:55, Ian Osgood wrote:
>
>>On July 07, 1999 at 02:02:18, Christophe Theron wrote:
>>
>>>On July 06, 1999 at 16:07:41, Ian Osgood wrote:
>>>
>>>>On July 06, 1999 at 15:31:06, Christophe Theron wrote:
>>>>
>>>>>On July 06, 1999 at 14:34:24, Ian Osgood wrote:
>>>>>
>>>>>>On July 05, 1999 at 15:12:55, Pierre Bourget wrote:
>>>>>>
>>>>...
>>>>>>>
>>>>>>>It would be interesting if someone can play a match of a dozen games between the
>>>>>>>Cosmos and the Sapphire II to see once and for all who is the best portable
>>>>>>>chess computer.Anyone willing to do that ?
>>>>>>>
>>>>>>>Pierre
>>>>>>
>>>>>>Great idea!
>>>>>>
>>>>>>This is easy to do for anyone who owns a Cosmos and has access to FICS (telnet
>>>>>>freechess.org 5000).
>>>>>>
>>>>>>Login and play SapphireII, an auto-playing Novag Sapphire II.  I try to keep it
>>>>>>available 24 hours a day, although I have been having problems with my internet
>>>>>>connection lately.
>>>>>
>>>>>Huh... How is it possible? You can't operate that manually, do you? And I
>>>>>thought the Sapphire was a standalone computer. How is it possible to have it
>>>>>connected to the Internet to play automatic games? Did you build a robot with an
>>>>>arm to operate the small thing and a video camera to look at the small
>>>>>display??? :)
>>>>
>>>>Many models of Novag computers came with a serial port, through which you can
>>>>send commands and receive results (such as the computer's move and thinking
>>>>analysis).  I wrote a WinBoard engine which translated WinBoard engine commands
>>>>into Novag serial commands, allowing my Sapphire II to play on the Internet or
>>>>from WinBoard! It is also handy to run test suites against the SapphireII
>>>>(similar to crafty's "test" command).  The engine also translates chess server
>>>>incremental time controls into appropriate thinking levels.
>>>>
>>>>SapphireII has played over 5000 automated games on FICS, maintaining a rating in
>>>>the range 2000-2200.  Many of the amateur chess authors here have appreciated
>>>>having a program of some sophistication (albeit slow) to spar against.
>>>>
>>>>BTW, the serial port is non-standard, requiring an adapter (called the Novag
>>>>Distributor) to translate signals to RS-232.  Most people don't buy the adapter
>>>>because it is an outrageous $70!  This  is also the method used to connect the
>>>>SapphireII to the Novag Universal Chess Board (sold as the Sapphire II DeLuxe).
>>>>
>>>>Come over to FICS and try it for yourself, Christophe!
>>>>
>>>>Ian
>>>
>>>
>>>Wow! Good work Ian!
>>>
>>>Unfortunately I will not have the time to challenge the SapphireII these days.
>>>However, if I do it one day I will try to run Tiger on a computer of the same
>>>class as the SapphireII processor. How fast is it? Is it 20MHz? In this case I
>>>can play with my 386sx 20MHz notebook to equalize chances...
>>>
>>>
>>>    Christophe
>>
>>The Sapphire II runs at 32 MHz (divided down to 16 MHz).
>
>What do you mean by "divided down to 16 MHz" ???
>
>
>
>>  It has a 120K position
>>opening book, and 128K RAM for hash tables (don't know how many positions that
>>is).  It reports between 3-5K nps over the course of a blitz game.  Quite good
>>for a portable, but no match for PC programs.
>
>I could give Tiger 128K hash table size on the 386sx 20MHz which is in fact a
>very bad machine for my program. Tiger is a true 32 bits program, so each time I
>get or send an integer to memory I need 2 memory cycles (the 386sx has a 16 bits
>memory data bus !!!). The funny thing is that I hardly never need my integers to
>be 32 bits long. Generally 16 bits integers (and even 8 bits integers) are
>enough!
>
>I can do only 700 nodes per seconds on this computer.
>
>I could use my very small opening book (7500 moves).
>
>
>>As a measure of its strength, here are some of its results against slower FICS
>>computers:
>>
>>oldman (crafty 486) 1-1=1
>>ezcape (crafty P83) 2-2
>>PoorGnu (gnuchess P120) 9-4=7
>>
>>SapphireII beats weakened computers, like TheComputer (CST level 1), Snafu, and
>>Wronskian; and loses to computers which run faster than 100 MHz.  With its tuned
>>Kittinger program, SapphireII also has an edge against amateur programs such as
>>PostModernist, Amateur, POWERHOUSE, JRCP, and HMChess.  SapphireII is
>>specifically weak against crafty with its excellent tactical extensions.
>
>Well... I'm almost sure that Tiger on my 386sx20 would have a bad time against
>the Sapphire...!
>
>
>    Christophe


I do appreciate having Sapphire to play against.  Since it's a known quantity, I
can test changes to Amateur easily by playing a few games, and noting the
results.  But I'd have to dispute your assertion that Sap has the edge against
my program, or the others you mentioned.  I think it averages out to about
3-1-1, usually.  But, as I said, I use fics mainly for experimenting, so results
aren't reliable.

Will



This page took 0 seconds to execute

Last modified: Thu, 15 Apr 21 08:11:13 -0700

Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.